██████╗ ██████╗ ██╗ ██╗ █████╗ ███╗ ██╗ ███╗ ███╗ █████╗ ██████╗ █████╗ ██████╗
██╔══██╗██╔═══██╗██║ ██║██╔══██╗████╗ ██║ ████╗ ████║██╔══██╗██╔══██╗██╔══██╗██╔══██╗
██████╔╝██║ ██║███████║███████║██╔██╗ ██║ ██╔████╔██║███████║██████╔╝███████║██████╔╝
██╔══██╗██║ ██║██╔══██║██╔══██║██║╚██╗██║ ██║╚██╔╝██║██╔══██║██╔══██╗██╔══██║██╔══██╗
██║ ██║╚██████╔╝██║ ██║██║ ██║██║ ╚████║ ██║ ╚═╝ ██║██║ ██║██║ ██║██║ ██║██║ ██║
╚═╝ ╚═╝ ╚═════╝ ╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═══╝ ╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝
> whoami
class Rohan:
role = "AI Engineer & Author"
school = "MS in Artificial Intelligence @ SUNY Buffalo"
based_in = "wherever curiosity takes me"
obsessed_by = ["neural networks", "deep learning", "computers", "general relativity", "quantum mechanics", "first-principles thinking", "the path to AGI"]
builds = ["AI products", "software apps", "books", "homelab infra"]
motto = "obsession > motivation"I treat AI, math, and code as tools to solve mysteries, find hidden patterns, and build cool things by trying out new stuff. I don't chase motivation — I run on obsession, ambition, and a refusal to settle for mediocre.
+ Deep-dive autopsies→ debugging and optimizing my latest deployed models
+ Homelab v2 → more self-hosted services, deeper devops
+ Research rabbit → obsessing over frontier AI breakthroughs
+ Writing → my next book: Echoes in the GradientSelf-published on Amazon. Fiction, non-fiction, philosophy, and one deep learning textbook.
| Title | Genre | |
|---|---|---|
| 📘 | Deep Learning Deciphered — neural networks explained through the lens of Sherlock Holmes | Technical |
| 🔍 | Grounds For Suspicion: A Julius Brandt Murder Mystery — a whodunit in the spirit of the classics | Mystery |
| 💱 | The Incentive Engine: Capitalism Works (Yes, Really) — three rivals, one collapsing reserve currency, the universe as a balance sheet | Sci-Fi |
| 🐢 | Achilles and the Tortoise: Dialogues on a Godless Universe — faith vs. science, in the spirit of classic philosophical dialogues | Philosophy |
📝 Cadence
Personalized research paper discovery platform. A "Spotify" for academia across multiple genres like CS, math, biology, economics, and so much more. This web app tracks your likes, saves and skips, and builds a highly accurate learned profile of your tastes and accordingly populates your feed with research papers you are more likely to enjoy reading, thus recommending you stuff according to what you like.
AI-powered nutrition app for Android. Check it out on Google Play Store now! Custom computer-vision models that identify 260+ foods in real time, barcode scanner, meal planner generating across 2M+ recipes daily, and full progress tracking. Built end-to-end under my startup Tensorbuilds — PyTorch models trained from scratch, React Native frontend, deployed on Oracle Cloud.
PyTorch · Computer Vision · React Native · Ubuntu
Personal homelab on bare-metal Proxmox. Self-hosted containers + VMs, reverse proxies, Cloudflare tunnels, SSL cert management, custom subdomain routing. A deep dive into the unglamorous side of devops.
Proxmox · Docker · Linux · Cloudflare · Nginx
Object detection with transfer learning — experiments across pretrained models plus a novel iterative fine-tuning approach for custom datasets. Full technical report in the repo.
PyTorch · YOLO-NAS · Transfer Learning
languages: [Python, C++]
ml_dl: [PyTorch, scikit-learn, OpenCV, HuggingFace]
ai_apps: [LangChain, RAG, Vector DBs, LLM fine-tuning]
mobile_web: [React]
infra: [Docker, Proxmox, Linux, Cloudflare]
data: [NumPy, Pandas, MongoDB]- ✍️ Blog — essays on cinema, travel, tech, and whatever I'm obsessing over this week
- 📺 YouTube — videos about what I do
- 🍳 Macros with Flavor — high-protein recipes (cookbook in the works)
> "My dream is to build superintelligent AI systems that can recursively improve themselves and usher us into an age of unstoppable technology singularity"
> end_of_transmission_